Results for “model”

arXiv (research papers) Yihan Hank Tang University of Pennsylvania SI

TrialDesignBench: Benchmarking Language Models for Phase-I Trial Design Against Clinical Knowledge

AI is used in much of the clinical trial pipeline. However, selecting the statistical design of a trial, a crucial step that determines how drug doses are tested on humans and evaluated, has seen almost no AI progress. LLMs appear to be a natural fit given their strengths in biomedical knowledge and statistical reasoning. Nevertheless, there is currently no appropriate way to evaluate such abilities because recent regulatory and methodological shifts, most notably FDA's…

Read at arXiv (research papers)

arXiv (research papers) Zhu Xiao SI

Beyond Sunny Skies: Using Large Language Models to Predict Urban Traffic in Adverse Weather

Urban traffic flow prediction is a cornerstone of intelligent transportation systems, yet accurate forecasting under adverse weather conditions remains a significant challenge. Existing data-driven models, while effective for routine patterns, often struggle with the sparse data and complex non-linear shifts associated with rare meteorological events. These methods typically treat environmental contexts as shallow numerical features, failing to capture the underlying…

Read at arXiv (research papers)

arXiv (research papers) Dezhen Lu Security

How Far Does Second-Order Linear Response Predict Quantization Mismatch? Pre-Registered Measurements on Five Language Models

Serving systems choose a weight format for each model and deployment. A low-cost screen predicts the resulting KL divergence from the unquantized model's Fisher information using $\tfrac12Δ^\top FΔ$, the expansion underlying Hessian-aware and second-order post-training quantizers. We assess the reliability of this screen in GPU batches with channel-wise int8, per-tensor fp8 and group-wise int4 on Qwen3-8B, Qwen3-4B, OLMo-2-7B, Qwen3-32B and Qwen3-30B-A3B. Relative…

Read at arXiv (research papers)

arXiv (research papers) Pujan Pant SI

Adapting Small Language Models to Romanized Nepali: Tokenizer Fertility, Reference Likelihood, and Generation Metrics

Romanized Nepali, Nepali written in Latin script, is common in digital communication but has no standard spelling and few resources, and it is unclear how reliably small open language models can be adapted to it. We adapt three 2-3B base models from different families, Llama-3.2-3B, Qwen2.5-3B, and Gemma-2-2B, with QLoRA on 9,000 instruction examples produced by deterministic transliteration of Nepali instruction data. Each model is trained with one shared recipe and three…

Read at arXiv (research papers)

The Verge Stevie Bonifield SI

Trump’s 60 Minutes interview touched on AI safety, threat to arrest an AI exec

Asked about AI executives' warnings about their models' safety on 60 Minutes, Trump said, "I think that the risk is not nearly as real as it- as they talk, and I think the risk can be managed, and it's managed." He claimed the DOJ is…

Read at The Verge

arXiv (research papers) Andrew Trask SI

Double Blind Evals: Resolving the Dual Confidentiality Dilemma in AI Safety Auditing

AI safety evaluations often involve sensitive prompts and proprietary models. Evaluators want to keep their evaluation prompts secret from model owners to prevent purposeful or incidental data leakage, ensuring longer usefulness of their benchmark. Model owners in turn wish to keep their proprietary model weights and inference code within their own secure infrastructure to protect their IP. We present a way forward for "double-blind evals", which provide cryptographic…

Read at arXiv (research papers)

The Decoder Maximilian Schreiner SI

Tencent's AI chief says AI bioweapons are coming, but closing off models isn't the answer

Tencent chief AI scientist Vinces Yao Shunyu warns that AI will soon be able to develop bioweapons and argues for a middle ground between open and closed models. While safety filters on open-source systems can be easily stripped through abliteration, closed models like Anthropic's Claude often block too aggressively. For its upcoming open-weight model Hy4, Tencent is filtering dangerous hacking knowledge directly from the training data. The article Tencent's AI chief says AI…

Read at The Decoder

MacRumors Joe Rossignol Gadgets

MacBook Pro With PROLED XDR Display: Details You Might Have Missed

The new MacBook Pro models with PROLED XDR displays began arriving to customers on Friday, and various details about the laptops continue to surface. Given these models are available with the same M5 Pro and M5 Max chips that power the models with mini-LED displays, performance was expected to be virtually identical across the board, and benchmark results confirm that is the case. Average scores for multi-core CPU performance in the Geekbench 7 database: 16-inch MacBook Pro…

Read at MacRumors

The Verge Jess Weatherbed SI

Disney and Adobe are building generative AI tools together

The Walt Disney Company has teamed up with Adobe to develop proprietary "artist-led" generative AI models to support Disney's production pipeline. The multiyear partnership was announced today at Adobe's MAX conference, and it will see…

Read at The Verge

arXiv (research papers) Victor Mattos Security

Covert System Identification Under Timing Constraints: A Static Linear Model and Its Detectability Implications

This paper studies covert system identification in cyber-physical systems, where an attacker infers plant dynamics while attempting to remain stealthy. We introduce a lightweight static linear model in which the time to execute a setpoint change is proportional to the commanded speed difference. On a Siemens PROFINET testbed this gain captures the drive's ramp response accurately but not its terminal overshoot. That omitted transient becomes the attacker's main exposure: on…

Read at arXiv (research papers)

MacRumors Tim Hardwick Gadgets

Some HomePod Mini 2 Buyers Returning Units Over Mac Speaker Issues

One of the features of Apple's new HomePod mini 2 speakers is being able to use them as your Mac's default speakers, but it seems to be failing to work as advertised for a number of users, with some owners already returning them, based on the latest MacRumors forum reports. OG HomePod minis in a stereo pair on a Mac (don't try this at home) The new HomePod mini 2 can be used as a default speaker on all Mac models with Wi-Fi 6E or later, according to Apple. A single HomePod…

Read at MacRumors

arXiv (research papers) Ayesha Habib Security

Text2NGAC: a multi-LLM pipeline for NGAC policy extraction with mutation-based verification

Access-control policies are written in natural language but enforced as formal models, and translating them into NIST Next Generation Access Control (NGAC) graphs remains unsolved: prior systems cover only subsets of the NGAC model, and none verify the graphs they extract. We present Text2NGAC, a pipeline of cooperating large language models that disambiguates each sentence before extraction, escalates uncertain sentences to a multi-model ensemble, and checks the resulting…

Read at arXiv (research papers)

arXiv (research papers) Mohammed Qaiss Security

ML Models Are Not Ciphers: Power Analysis for Weight Extraction Is Harder Than You Think

Can power-analysis techniques reveal secret weights of machine-learning models? Multiple recent works demonstrate proof-of-concept attacks that can partially recover such weights. To that end, these works build on techniques from cryptographic engineering, treating the model as a ``cipher'' and its secret weights as keys. We question the basic assumption behind such straightforward adaptations of cryptographic attacks and highlight the challenges arising when applying past…

Read at arXiv (research papers)

arXiv (research papers) Julien Delaunay SI

Can the Confidence and Decisions of System 1 Models Be Trusted? An Explainability and Perturbation Audit of Jev-Style Decision Models

"System 1" decision models (pioneered by TypeSafe AI's Jev) replace autoregressive text generation with a single forward pass returning typed probability distributions over declared schemas (Choice, Score, Noul). While deployed for low-latency routing and automated triage, their statistical calibration and post-hoc explainability remain uncharacterized. We introduce an open, deterministic audit benchmark evaluating System 1 decision models across calibration under…

Read at arXiv (research papers)

9to5Mac Ryan Christoffel Gadgets

Leak reveals details for more iPhone 18 models launching early 2027

Apple is rumored to have three new iPhone models launching in early 2027: iPhone 18, iPhone 18e, and iPhone Air 2. Here’s what to expect from the iPhone 18 models per the latest leak.

Read at 9to5Mac

SemiWiki Jonah McLeod SI

Who Owns the Voice in Your Head?

On May 13, 2024, OpenAI demonstrated GPT-4o, a model that could carry on a spoken conversation with the timing and inflection of a person. One of its voices, called Sky, was warm, quick and more than a little flirtatious. After the demo, Sam Altman posted a single word on X: “her.” Nobody in the industry needed the reference…

Read at SemiWiki

arXiv (research papers) Sameer Deshmukh

To SIMD or Not to SIMD: Comparing RVV vs. CGRA SoCs for Non-Linear Functions in AI

Non-linear functions are an important class of operations in modern large language models. However, chip vendors building for AI have so far relied on traditional vector or scalar processing units for non-linear functions. The performance of these units has not kept pace with matrix multiplication units (MMUs), which can be found in both high-performance and edge-computing devices from almost every vendor. Reconfigurable units such as FPGAs and CGRAs have been proposed in…

Read at arXiv (research papers)

GEN (Genetic Engineering & Biotechnology News) Kathy Vuksanaj

New AI Models Are Turbo-Charging Protein Design

Computer-designed protein therapeutics move from proof-of-concept to wet-lab validation and even clinical trials as the field heats up.

Read at GEN (Genetic Engineering & Biotechnology News)

arXiv (research papers) Dario Pasquini SI

Say Shibboleth: Fingerprinting the LLMs and Harnesses Behind Agentic Attacks

We study the fingerprinting of agentic cyberattacks. Our objective is to reliably infer information about the attacker's configuration, such as LLM version and harness, from telemetry. To achieve this, we introduce a technique we call shibboleth traps: modular honeypot services designed to elicit discriminative behaviors from agentic attackers. The traps make informative interactions a condition for obtaining resources the agent seeks, eliciting model declarations and…

Read at arXiv (research papers)

arXiv (research papers) Dario Pasquini SI

Say Shibboleth: Fingerprinting the LLMs and Harnesses Behind Agentic Attacks

We study the fingerprinting of agentic cyberattacks. Our objective is to reliably infer information about the attacker's configuration, such as LLM version and harness, from telemetry. To achieve this, we introduce a technique we call shibboleth traps: modular honeypot services designed to elicit discriminative behaviors from agentic attackers. The traps make informative interactions a condition for obtaining resources the agent seeks, eliciting model declarations and…

Read at arXiv (research papers)

arXiv (research papers) Jeremy Canale Security

Detecting Shadow AI: Unsanctioned LLM Agents on Enterprise Endpoints: A Taxonomy, an Open Detector and a Controlled Evaluation

Shadow IT, the use of technology that the IT organisation has not approved, is a long-standing governance problem. We argue that autonomous AI agents can make it grow much faster: agents make building tools nearly free, act with their user's privileges, and leave behind scripts, automations and MCP servers that are themselves unsanctioned. A simple compounding model shows that, under its assumptions, shadow IT grows exponentially once agentic artifacts reproduce faster than…

Read at arXiv (research papers)

TechCabal Adonijah Ndege Startups

Money Fellows took an old savings habit and built a business the hard way

Money Fellows built its model around digitising an old saving habit in Egypt.

Read at TechCabal

OpenAI News SI

Control Room: A central system to track and stop model workloads across our research infrastructure

Learn how OpenAI’s Control Room tracks model workloads across our research infrastructure and can rapidly stop them when safety or security concerns arise.

Read at OpenAI News

MIT Technology Review MIT Technology Review Insights SI

Innovating with AI in the special chemicals industry

As companies look to accelerate innovation with artificial intelligence, cloud computing, and advanced modeling, the primary challenge is connecting new technologies to reliable data, scientific expertise, and the practical realities of…

Read at MIT Technology Review

arXiv (research papers) Michele Pizzochero Tech policy

Do Machine Learning Models Predict?

Machine learning (ML) models are said to predict. Philosophers draw a line between accommodation and prediction, depending on whether or not the evidence a theory succeeds on was used in constructing it. Does this distinction survive when a theory is replaced by an ML model? We argue so, yet the dividing line becomes fuzzy. Whether an output of an ML model counts as an accommodation or as a prediction hinges upon the relationship between the patterns the input instantiate…

Read at arXiv (research papers)

Inc42 Palak Sharma Startups

Decoding BlackBuck’s Core Trucking Model And Its New Engines

A truck carrying goods across India spends several days or even weeks on the road, but it’s not just the…

Read at Inc42

Inc42 Palak Sharma Startups

Decoding BlackBuck’s Core Trucking Model And Its New Engines

A truck carrying goods across India spends several days or even weeks on the road, but it’s not just the…

Read at Inc42

Crunchbase News Guest Author Startups

Paper Unicorns Are Cheap. Building A $1B Revenue AI Company Takes A Multimarket Cap Table

AI startups should be judged by durable revenue, full-price renewals and healthy margins after delivery costs, with embedded customer workflows providing stronger defensibility as model access gets cheaper, rather than their paper…

Read at Crunchbase News

Buenos Aires Times Ariel Maciel Geopolitics

Pope Leo challenges Milei's vision for Argentina's economy

In front of President Javier Milei and his Cabinet, Vatican leader raises concerns over plight of pensioners and exodus of young people abroad; Catholic leader defends education funding and questioned an austerity model that lacks social inclusion. Leer más

Read at Buenos Aires Times

arXiv (research papers) Melih Coşğun Security

SNAPRAG: Cost-Efficient Knowledge Extraction Attack on RAG Systems via Adversarial Perturbations

Retrieval-Augmented Generation (RAG) systems improve Large Language Model (LLM) responses by incorporating information from external knowledge bases. However, recent studies have shown that attackers can extract substantial portions of private knowledge bases from RAG systems through repeated querying. Yet, state-of-the-art extraction attacks either rely heavily on external LLMs to generate adversarial queries or require a large number of queries. In this paper, we propose…

Read at arXiv (research papers)

Older stories